CN103441952A - Network data package processing method based on multi-core or many-core embedded processor - Google Patents
Network data package processing method based on multi-core or many-core embedded processor Download PDFInfo
- Publication number
- CN103441952A CN103441952A CN2013103656075A CN201310365607A CN103441952A CN 103441952 A CN103441952 A CN 103441952A CN 2013103656075 A CN2013103656075 A CN 2013103656075A CN 201310365607 A CN201310365607 A CN 201310365607A CN 103441952 A CN103441952 A CN 103441952A
- Authority
- CN
- China
- Prior art keywords
- core
- data
- packet
- processed
- multinuclear
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Abstract
The invention discloses a network data package processing method based on a multi-core or many-core embedded processor. The method includes the steps that (1) cores in the multi-core or many-core embedded processor are divided into data receiving cores and data processing cores, the data receiving cores perform buffer allocation on a shared memory region, the data processing cores perform distribution strategy registration on the data receiving cores, and the characteristics of data packages to be processed by each data processing core are determined; (2) the data receiving cores store the data packages into corresponding buffers according to data package header information received from network hardware; (3) the data receiving cores send messages to the data processing cores according to the distribution strategy registered by the data processing cores, the data processing cores read the data packages in the buffers for processing according to the messages and send processing state information to the data receiving cores.
Description
Technical field
The present invention relates to multinuclear or many core flush bonding processors and network service process field, relate in particular to a kind of processing method of network data packets based on multinuclear or many core flush bonding processors.
Background technology
Traditional single core processor tupe is the single task serial process, can only increase dominant frequency in order to improve processing speed, and the raising of dominant frequency is conditional, and has increased power consumption.Multinuclear or many-core processor solve these problems well, and multinuclear or many-core processor adopt parallel processing mode, and multi-task parallel is processed, and can improve processing speed in the situation that hang down dominant frequency, and reduce power consumption.And along with the increase day by day of bandwidth, the explosive growth of network data amount.The rise of Internet of Things heat has aggravated the demand to the network data fast processing especially, numerous data in network all require very strong real-time simultaneously, for the smooth transport that guarantees data and the flowed fluctuation while reducing transfer of data, how software systems effectively utilize its framework advantage to carry out the high-speed receiving of network packet and the focus that distribution becomes research already.
The patent of invention that application number is CN201010289667.X discloses a kind of message forwarding method and system based on multicore architecture, this invention is by obtaining the packet in the Internet, and when packet is the tunnel links packet, it is carried out to decapsulation, obtain the IP packet, adopt hash algorithm to carry out hash algorithm to the five-tuple of IP packet, again according to calculate cryptographic Hash and the quantity of Hash table size and CPU, cryptographic Hash is tagged in above-mentioned IP packet, then is sent in described CPU.
Although the foregoing invention patent is a kind of file transmitting method about multicore architecture, but be in fact only a kind of forwarding of message, do not relate to data receiver and distribution in multinuclear or many core embedded systems, the strategy of more not mentioned Data dissemination, and for multinuclear or many core flush bonding processors, significant for the research of the receiving management of its network packet and point-score processing method.
Summary of the invention
The invention provides the receiving management of network packet in a kind of multinuclear or many core embedded processor systems and the distribution policy method of packet, can effectively realize accurately fast reception and the distribution of network packet.
A kind of processing method of network data packets based on multinuclear or many core flush bonding processors comprises the following steps:
1) core in multinuclear or many core flush bonding processors is divided into to data receiver core and data and processes core, described data receiver is checked the shared drive zone and is carried out Buffer allocation, and data are processed core and carried out the distribution policy registration to described data receiver core, determine each data processing core packet feature to be dealt with;
2) data receiver core, according to the packet head information received from the network hardware, is deposited packet to corresponding buffering area;
3) data receiver core processes according to data the distribution policy that core is registered, process core to described data and send message, described data are processed core according to described message, and the packet read in buffering area is processed, and to described data receiver core transmission processing state information.
Described multinuclear or many core flush bonding processors are isomorphism or heterogeneous processor, and the quantity of described core is at least 8, and wherein, the quantity of described data receiver core is one or more.
The destination data that the allocation manager, data that data receiver core is responsible for network data reception buffer zone in the shared drive zone processed the management of core registration distribution policy and pending packet is processed the judgement of core, data receiver core also is responsible for configuration and the data interaction of network hardware equipment, comprise the mode of operation of configuration network interface card hardware and obtain its state parameter etc., data interaction refers to the reception of network data; Data are processed core and are carried out the registration of distribution policy according to practical application request to data receiver core, and extract the packet left in buffering area and processed.
The number that the number of data receiver core and data are processed core all can be configured according to practical application.For example, choose 8 cores and carry out simple image binaryzation processing, data receiver core only receives complete piece image, then distributes and get final product, and therefore chooses data and receives core and just can complete aforesaid operations; When image is processed, can choose 7 remaining cores and carry out binary conversion treatment as data processing core.Simultaneously, data receiver is endorsed according to composite factors such as processor structure, network hardware equipment position, core positions and is set, it is the processing core nearest apart from the network hardware interface position that general data receives core, data are processed the number of core and position and are selected determine according to application demand and data receiver core, and data processing core can be set up on their own by the user fully.
In step 1) in, the size of described buffering area is 256 bytes, 512 bytes, 1024 bytes, 1536 bytes, 2048 bytes, 4096 bytes or 10240 bytes, the quantity of described buffering area is a plurality of, and described data are processed the physical address that nuclear energy is checked each buffering area.
In a plurality of buffering areas, the quantity of the buffering area of every kind of size is configured according to the packet feature, the physical address of all buffering areas can be processed to verify by data and see simultaneously, that is to say that these buffering areas all share for all processing core, each buffering area is except the packet content of the appointment size of depositing reception, also comprised and this Packet Generation will have been processed to the information of core to which data, the data of this packet being processed are processed the information of core.The specifying information of each buffering area is as shown in table 1:
Data buffering structure in table 1 data receiver core
In table 1, in its structure of buffering area of different sizes, buffer data varies in size, the size of the processing core sign of buffering area varies in size and difference according to the processing core of used flush bonding processor, the time that sends to data to process core message refers to when data receiver core judges which data these record data send to process core, the timestamp added at this data buffering afterbody.
Described step 2), in, described packet feature comprises type of data packet, hardware header information type and port information, the packet characteristics determined process its network data data process core.
In step 1) in, each data is processed core can only register a distribution policy, and same distribution policy can be processed core by a plurality of data simultaneously and register.
Data receiver core is for to management and the distribution of distribution policy, can each data of dynamic-configuration processes the distribution policy of core, and distribution policy number is altogether processed check figure and determines according to data, and each core can only be registered a distribution policy.Data are processed core and are responsible for carrying out the distribution policy registration by certain strategy foundation or application demand to data receiver core, in registration process, specify and send to the packet feature on this registration core.
Described distribution policy comprises data pack protocol strategy, packet MAC Address strategy, IP the address Policy and packet port policy.
In the present invention, data process distribution policy information that core sends to data receiver core as 2 of following tables not:
Table 2 data are processed the tactful content of making a determined effort of core
In table 2, data processing core carries out to data receiver core in the information of distribution policy registration, and the core decision is processed in determining by data of each distribution policy, can be comprehensive distribution policy, can be also the configuration of single distribution policy.Shared sign is also registered identical strategy if there are other data to process cores after meaning that this data processing core receives the packet that meets its distribution policy, whether is ready that processing cores with other data shares.
When each data processing core carries out the distribution policy registration, data receiver core judges whether the distribution policy of this data processing core request is that other data are processed the registered distribution policies of cores, if registered and need to have been shared with other core, the shared flag bit of each data being processed to core is upgraded.
After data receiver core receives the distribution policy of data processing core, can carry out storage administration in data receiver core inside to each distribution policy, in storage administration, each data is processed the content of distribution policy in above table 2 of core, also comprises following two contents:
1): whether the distribution policy that this data processing core is registered has identical flag bit with the distribution policy of other core registration.When any one data processing core carries out the distribution policy registration to data receiver core, data receiver core can be processed the registered distribution policy of core to data and be judged, more new data is processed this flag bit of core.
2): the number of data packets that this network data processing core had received.This processes for each data of Real-time Obtaining the Data Dynamic that core receives, and can be used for carrying out the subsequent network analysis.
Data receiver is checked network data and is distributed when judgement, and at first according to the packet feature received, the registration distribution policy that utilizes each data to process core judged, traversal finds the data that meet this packet feature to process core.If but the registration of data processing core is sharing policy, directly by the processing core of the buffering area in this buffer structure described in table 1 sign correspondence position position, upgrade the corresponding data that send to and process the time of core message, then send a message to data processing core and get final product.Described message comprises address and the buffer size of packet in buffering area.
After data receiver core has determined that the data of current data packet are processed core, adopt the message sending strategy of corresponding multinuclear or many core flush bonding processors to carry out the message transmission to each data processing core, message comprises the buffer address of pending data, the contents such as buffer size.After data processing core is disposed buffering, this processing of the processing core of buffering area sign neutralization need to be checked to the bit zero clearing of answering.
When if the distribution policy that a plurality of data processing core is registered is identical, data receiver core is processed processing data packets state and number of data packets in core according to each data, pending packet is carried out to weight allocation calculating, according to weight, process core to a plurality of data packet delivery and process simultaneously.
For example, can use analytic hierarchy process (AHP) to carry out decision-making, process the processing data packets speed of core, the number of pending packet by the analysis-by-synthesis data, and the multiple performance relevant parameters such as size of pending packet, structure compares judgment matrix in twos, by with method, minimum clamp horn cupping or characteristic vector method etc., asking for the weight that different pieces of information is processed core, by comparing weighted value, judge that pending packet is by the data processing core be sent to.
If data are processed the core registration, be that non-sharing policy and a plurality of data are when processing core and having registered the same distribution policy, need according to time, the treatment state of packet and the number of data packets that this core has received of processing the last packet of core transmission to data, comprehensively to be judged simultaneously, carry out the selection that destination data is processed core, can be processed by a plurality of data processing core simultaneously.
Beneficial effect of the present invention is:
(1) the network data cache management that the present invention proposes and the autgmentability of network data tactical management are very strong, go for most multinuclear or many core flush bonding processors, can meet most network packet detecting application demand.
(2) processing that in the present invention, network data processing is checked data has realized zero-copy, is conducive to the lifting of processing speed.And can be a plurality of data process and check same network packet and carry out the concurrent collaborative processing.
The accompanying drawing explanation
Fig. 1 is entire block diagram of the present invention;
Fig. 2 is registration and the flow chart of data processing figure that in the present invention, data are processed core;
The flow chart of data processing figure that Fig. 3 is data receiver core in the present invention;
The Snort intruding detection system block diagram that Fig. 4 is parallelization in the present invention;
The basic flow sheet of Fig. 5 for using the present invention to carry out Snort intrusion detection realization.
Embodiment
Below by embodiment, the present invention will be described in detail.
The present invention is a kind of network packet receiving management and distribution processing method based on multinuclear or many core flush bonding processors.Figure 1 shows that entire block diagram of the present invention, be divided into the network hardware, data receiver core, four modules of data processing core and shared drive zone.In figure: network hardware equipment is put in 1 expression data receiver caryogamy; 2 mean that data receiver core carries out network data and receives distribution and the management of depositing buffering; 3 mean that data processing core carries out the strategy registration to data receiver core; 4 mean that data receiver core is after the packet head information received from hardware device, find suitable buffering area from internal memory, transfer the destination of entire data using its address as network hardware equipment; 5 mean that network hardware equipment sends message to data receiver core and completes to show data receiver; 6 mean that the data receiver cores process core according to the distribution policy configuration information to corresponding data and send message, comprise the address of data buffer zone in message, show current have complete data received and etc. pending; The data that 7 expression data processing cores read in the corresponding address buffering area are specifically processed.
In the present invention, multinuclear or many core flush bonding processors are used the Tile64 polycaryon processor.This processor is integrated 64 tile on single-chip, with the power consumption of the highest 23W, can provide the 192BOPs disposal ability of (1,000,000,000 operations of per second), the built-in fields such as main network-oriented processing, multimedia, information security, radio communication.Tile64 is a novel polycaryon processor, has solved high handling property, the flexible problem of programmability and power supply power consumption simultaneously, and has erected feasible bridge between the performance that can bring for the growth of network application demand and existing processor.
In the present invention, the main process of data manipulation is as follows: data receiver core is at first to the network hardware equipment Transfer Parameters, and configuration network bindings equipment, mainly refer to the configuration of network interface card hardware effort pattern; After hardware device configuration, data receiver is exclusive to be checked in the shared drive zone network data and receives and deposit buffering area and specifically distribute, and according to actual application demand, it is assigned as to the multiple buffering areas of different sizes, and it is managed; Subsequently, data are processed core and are carried out the distribution policy registration according to certain strategy foundation or application demand to data receiver core, can specify the packet feature sent on this registration core in this process, comprise type of data packet, hardware header information type, port and other information etc.; After the strategy registration, data receiver core, according to the information that receives the packet head, is found suitable buffering, the destination of carrying out the partial data transmission using its address as network hardware equipment from internal memory; After network hardware equipment completes the data transmission, send message to data receiver core and show that data receiver completes; Data receiver core receive network hardware equipment complete message after, the distribution policy configuration information of processing core according to each data is processed core to corresponding data and is sent message (comprising the buffer address that receives pending data in message), shows the current event that has; Final data is processed the message of core according to data receiver core, and the data that read in the corresponding address buffering area are specifically processed.
In the present invention, be divided into data receiver core and data processing core according to function.Wherein the concrete flow chart of data processing of data processing core as shown in Figure 2.At first data are processed core needs communicating by letter of foundation and data receiver core, data are processed core and could be sent message to data receiver core on this basis, the registration distribution policy, after distribution policy succeeds in registration, data are processed core and are entered wait state, until sending message to it, data receiver core shows the current event that has, pending data buffering is arranged, the message that network data processing core sends according to data receiver core, obtaining data buffering is processed, after being disposed, send message to data receiver core again and show that data are disposed, above-mentioned is once the data handling procedure that complete data are processed core.
And the handling process of data receiver core as shown in Figure 3.Data receiver core, after startup, at first needs the configuration network bindings module, comprises the configuration to network interface card hardware effort pattern, then according to practical application request, buffering is assigned as to 256 bytes, 512 bytes, 1024 bytes, 1536 bytes, 2048 bytes, 4096 bytes and 10240 bytes.Then data receiver core enters the state waited for the arrival of news.If receive the registration message of processing core from data, according to message request, the distribution policy of location registration process core is upgraded registered content simultaneously.When any one data processing core carries out the distribution policy registration to data receiver core, data receiver core all can be processed the registered distribution policy of core to data and be judged, upgrades the flag bit that each processes core.If receive the information about the packet head from network hardware equipment, after it is analyzed, carry out the strategy judgement, choose suitable buffering from buffering area, carry out the DMA configuration to network hardware equipment simultaneously.If receive the message completed about data packet transmission from network hardware equipment, at first upgrade buffering area and distribution policy entry, then to data, process core and send message, show the current pending packet that has.
On the basis of above-mentioned network packet receiving management based on multinuclear or many core flush bonding processors and distribution processing method, can realize the Snort intrusion detection of parallelization, its system block diagram is as shown in Figure 4.Only by simply copying, the Snort program is operated on a plurality of processing core, for the restriction of data dependence relation, need to carry out following two special processings to program:
1, eliminate global data and rely on, each data is processed on core Snort is carried out as a process, and the data privacy feature of each process can solve the global data Dependence Problem.Simultaneously using Snort as process independently and data process core and carry out map bindings, this is the precondition that obtains parallel capture storehouse Libpcap Data dissemination.
2, eliminate file resource and rely on, log hereof in the Snort program operation process, when a plurality of data are processed and checked identical file and operated, easily bring the serialization problem.Process core according to data and have the uniquely identified characteristics, the sign of journal file and data processing core is bound, each Snort can have journal file separately like this, thereby has solved the problem that file resource relies on.
Intrusion detection process prescription based on the Snort system architecture is as follows: at first, the network packet that network is produced when being invaded is excavated, and finds out the regular characteristics with the network packet of attacking character; Then, will meet these regular primitive network packets as detecting the knowledge base of engine in intruding detection system; Finally, in real time the express network data are carried out to rule match, once coupling correctly just thinks that network intrusions produces, and notifies output module to process accordingly at once.Carry out number of times according to various piece and the system operation can be divided into to two stages.The initialization that first stage is system, after preparation completes, the first stage carry out to finish, unless the injection of new regulation is arranged or the phenomenon such as abnormal occurs, the first stage can not carry out again.At this one-phase, system mainly completes the work such as the structure generation, Network Data Capturing initialization of global data initialization, rule file parsing, plug-in registration, rule tree.Second stage is that system detects in real time, along with the arrival circulation of new data is carried out.At this one-phase, complete to the main cyclicity of system the work such as reception, Data Analysis, preliminary treatment, intrusion behavior detection and the output of reporting to the police of network packet.The basic procedure of system operation as shown in Figure 5.
The Snort system, after the first stage completes, enters second stage circular wait handle packet.The system finishing condition mainly contains two: the one, and the user has specified and has detected bag quantity, and resolved detection can be carried out along with the reception circulation of new network data, until detected the packet specified number; The 2nd, the user does not specify the detection data bulk, and resolved detection module acquiescence is processed by the individual data bag, until stop after the generation of system finishing event.After having detected a packet, all can be checked signal event at every turn, be carried out the operations such as renewal judgement of new regulation, in order to respond in time demand.
Through above-mentioned realization, can, by the Snort parallel running on multinuclear or many core embedded platforms, not only can promote the performance of detection system, and make system there is very strong extensibility.
Claims (10)
1. the processing method of network data packets based on multinuclear or many core flush bonding processors, is characterized in that, comprising:
1) core in multinuclear or many core flush bonding processors is divided into to data receiver core and data and processes core, described data receiver is checked the shared drive zone and is carried out Buffer allocation, and data are processed core and carried out the distribution policy registration to described data receiver core, determine each data processing core packet feature to be dealt with;
2) data receiver core, according to the packet head information received from the network hardware, is deposited packet to corresponding buffering area;
3) data receiver core processes according to data the distribution policy that core is registered, process core to described data and send message, described data are processed core according to described message, and the packet read in buffering area is processed, and to described data receiver core transmission processing state information.
2. the processing method of network data packets based on multinuclear or many core flush bonding processors as claimed in claim 1, it is characterized in that, described multinuclear or many core flush bonding processors are isomorphism or heterogeneous processor, the quantity of described core is at least 8, wherein, the quantity of described data receiver core is one or more.
3. the processing method of network data packets based on multinuclear or many core flush bonding processors as claimed in claim 2, it is characterized in that, in step 1) in, the size of described buffering area is 256 bytes, 512 bytes, 1024 bytes, 1536 bytes, 2048 bytes, 4096 bytes or 10240 bytes.
4. the processing method of network data packets based on multinuclear or many core flush bonding processors as claimed in claim 3, is characterized in that, the quantity of described buffering area is a plurality of, in step 3) in, described data are processed the physical address that nuclear energy is checked each buffering area.
5. the processing method of network data packets based on multinuclear or many core flush bonding processors as claimed in claim 1, is characterized in that described step 2) in, described packet feature comprises type of data packet, hardware header information type and port information.
6. the processing method of network data packets based on multinuclear or many core flush bonding processors as claimed in claim 5, it is characterized in that, in step 1) in, each data is processed core can only register a distribution policy, and same distribution policy can be processed core by a plurality of data simultaneously and register.
7. the processing method of network data packets based on multinuclear or many core flush bonding processors as claimed in claim 6, is characterized in that, described distribution policy comprises data pack protocol strategy, packet MAC Address strategy, IP the address Policy and packet port policy.
8. the processing method of network data packets based on multinuclear or many core flush bonding processors as claimed in claim 7, it is characterized in that, when each data processing core carries out the distribution policy registration, data receiver core judges whether the distribution policy of this data processing core request is that other data are processed the registered distribution policies of cores, if registered and need to process core with other data and be shared, the shared flag bit of each data being processed to core is upgraded.
9. the processing method of network data packets based on multinuclear or many core flush bonding processors as claimed in claim 8, is characterized in that, described message comprises address and the buffer size of packet in buffering area.
10. the processing method of network data packets based on multinuclear or many core flush bonding processors as claimed in claim 9, it is characterized in that, when if the distribution policy that a plurality of data processing core is registered is identical, in described step 3) in, data receiver core is processed processing data packets state and number of data packets in core according to each data, pending packet is carried out to weight allocation calculating, according to weight, process core to a plurality of data packet delivery and process simultaneously.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310365607.5A CN103441952B (en) | 2013-08-20 | 2013-08-20 | Based on multinuclear or the processing method of network data packets of many-core flush bonding processor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310365607.5A CN103441952B (en) | 2013-08-20 | 2013-08-20 | Based on multinuclear or the processing method of network data packets of many-core flush bonding processor |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103441952A true CN103441952A (en) | 2013-12-11 |
| CN103441952B CN103441952B (en) | 2016-10-05 |
Family
ID=49695618
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310365607.5A Active CN103441952B (en) | 2013-08-20 | 2013-08-20 | Based on multinuclear or the processing method of network data packets of many-core flush bonding processor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103441952B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104639460A (en) * | 2015-02-03 | 2015-05-20 | 重庆邮电大学 | High-speed network data packet parallel receiving method based on many-core processor |
| WO2016041447A1 (en) * | 2014-09-15 | 2016-03-24 | 华为技术有限公司 | Multicore processor chip-based data processing method, device, and system |
| CN105991755A (en) * | 2015-05-21 | 2016-10-05 | 杭州迪普科技有限公司 | Service message distribution method and service message distribution device |
| CN112650705A (en) * | 2020-12-31 | 2021-04-13 | 清华大学 | Routing control method and artificial intelligence processor |
| CN113138711A (en) * | 2020-01-20 | 2021-07-20 | 北京希姆计算科技有限公司 | Storage management device and chip |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101562544A (en) * | 2008-04-17 | 2009-10-21 | 浩通科技(深圳)有限公司 | Data packet generator and data packet generating method |
| US20110153982A1 (en) * | 2009-12-21 | 2011-06-23 | Bbn Technologies Corp. | Systems and methods for collecting data from multiple core processors |
| CN102301664A (en) * | 2011-07-20 | 2011-12-28 | 华为技术有限公司 | Method and device for dispatching streams of multicore processor |
| CN103166842A (en) * | 2011-12-13 | 2013-06-19 | 百度在线网络技术（北京）有限公司 | Message generation method based on multi-core platform and generation device |
-
2013
- 2013-08-20 CN CN201310365607.5A patent/CN103441952B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101562544A (en) * | 2008-04-17 | 2009-10-21 | 浩通科技(深圳)有限公司 | Data packet generator and data packet generating method |
| US20110153982A1 (en) * | 2009-12-21 | 2011-06-23 | Bbn Technologies Corp. | Systems and methods for collecting data from multiple core processors |
| CN102301664A (en) * | 2011-07-20 | 2011-12-28 | 华为技术有限公司 | Method and device for dispatching streams of multicore processor |
| CN103166842A (en) * | 2011-12-13 | 2013-06-19 | 百度在线网络技术（北京）有限公司 | Message generation method based on multi-core platform and generation device |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016041447A1 (en) * | 2014-09-15 | 2016-03-24 | 华为技术有限公司 | Multicore processor chip-based data processing method, device, and system |
| CN105589829A (en) * | 2014-09-15 | 2016-05-18 | 华为技术有限公司 | Data processing method based on multi-core processor chip, device and system |
| CN104639460A (en) * | 2015-02-03 | 2015-05-20 | 重庆邮电大学 | High-speed network data packet parallel receiving method based on many-core processor |
| CN105991755A (en) * | 2015-05-21 | 2016-10-05 | 杭州迪普科技有限公司 | Service message distribution method and service message distribution device |
| CN105991755B (en) * | 2015-05-21 | 2019-03-15 | 杭州迪普科技股份有限公司 | Service message distribution method and device |
| CN113138711A (en) * | 2020-01-20 | 2021-07-20 | 北京希姆计算科技有限公司 | Storage management device and chip |
| CN113138711B (en) * | 2020-01-20 | 2023-11-17 | 北京希姆计算科技有限公司 | Storage management device and chip |
| CN112650705A (en) * | 2020-12-31 | 2021-04-13 | 清华大学 | Routing control method and artificial intelligence processor |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103441952B (en) | 2016-10-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102904729B (en) | The intelligent acceleration network card of more applications is supported according to agreement, port shunt | |
| CN1206600C (en) | Full distribution type aggregation network servicer system | |
| US10079740B2 (en) | Packet capture engine for commodity network interface cards in high-speed networks | |
| CN103441952A (en) | Network data package processing method based on multi-core or many-core embedded processor | |
| US20190123894A1 (en) | Programmable hardware based data encryption and decryption systems and methods | |
| WO2015018188A1 (en) | Deep packet inspection method, device, and coprocessor | |
| CN101667144A (en) | Virtual machine communication method based on shared memory | |
| CN101729573B (en) | Dynamic load balancing method of network intrusion detection | |
| CN101997924A (en) | Cloud storage file transfer protocol (CFTP) | |
| EP3633948B1 (en) | Anti-attack method and device for server | |
| CN103049336A (en) | Hash-based network card soft interrupt and load balancing method | |
| CN105282169A (en) | DDoS attack warning method and system based on SDN controller threshold | |
| CN103581274B (en) | Message forwarding method and device in stacking system | |
| CN103875214A (en) | Intelligent phy with security detection for ethernet networks | |
| CN103491535B (en) | The general approximate enquiring method of secret protection of facing sensing device network | |
| KR101679573B1 (en) | Method and apparatus for service traffic security using dimm channel distribution multicore processing system | |
| CN101039310A (en) | Link sharing service apparatus and communication method thereof | |
| US9225999B2 (en) | Virtualized security processor | |
| CN106331184A (en) | Big data distribution method and distribution platform based on internet | |
| CN108965148A (en) | A kind of processor and message processing method | |
| CN102833336A (en) | Data sub-packet processing method in separate distributed information acquisition and concurrent processing system | |
| CN104639460A (en) | High-speed network data packet parallel receiving method based on many-core processor | |
| CN106992987A (en) | A kind of information transmission equipment and method based on USB | |
| CN102404211A (en) | Method and device for realizing load balancing of processors under AMP framework | |
| KR20120121668A (en) | High Performance System and Method for Blocking Harmful Sites Access on the basis of Network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210114 Address after: 311200 room b1-3-034, No. 198, Qidi Road, economic and Technological Development Zone, Xiaoshan District, Hangzhou City, Zhejiang Province Patentee after: Hangzhou purevision Technology Co.,Ltd. Address before: 710071 Xi'an Electronic and Science University, 2 Taibai South Road, Shaanxi, Xi'an Patentee before: XIDIAN University |